Download The closest extrasolar planet: A giant planet around the M4 dwarf Gl

arXiv:astro-ph/9808026v2 18 Aug 1998
1 2
1
A&A manuscript no.
(will be inserted by hand later)
ASTRONOMY
AND
ASTROPHYSICS
Your thesaurus codes are:
06 (03.20.7; 08.12.2; 08.12.1; 07.19.2)
The closest extrasolar planet
A giant planet around the M4 dwarf Gl 876
X. Delfosse12 , T. Forveille2 , M. Mayor1 , C. Perrier2 , D. Naef1 , and D. Queloz31
1
2
3
Observatoire de Genève, 51 Ch des Maillettes, 1290 Sauverny, Switzerland
Observatoire de Grenoble, 414 rue de la Piscine, Domaine Universitaire de St Martin d’Hères, F-38041 Grenoble, France
Jet Propulsion Laboratory, Mail Stop 306-473, 4800 Oak Grove Drive, Pasadena, CA 91109, USA
Received ; Accepted
Abstract. Precise radial velocity observations of the
nearby M4 dwarf Gl 876 with the Observatoire de Haute
Provence 1.93 m telescope and the new 1.20 m swiss telescope at la Silla indicate the presence of a jovian mass
companion to this star. The orbital fit to the data gives
a period of 60.8 days, a velocity amplitude of 246m.s−1
and an eccentricity of 0.34. Assuming that Gl 876 has a
mass of 0.3 M ⊙ , the mass function implies a mass for the
companion of 2/sin i Jupiter masses.
Key words: giant planet formation – extrasolar planets
– giant planets – M dwarf stars
1. Introduction
The still recent discovery of the first extrasolar planet,
around 51 Peg (Mayor & Queloz, 1995), has since been
followed by many more. The count currently runs to 11
very low mass companions (Marcy & Butler, 1998; Queloz,
1999), with minimum masses (M sin i) which range between 0.5 and 10 times the mass of Jupiter. Asides from
their Jupiter-like masses, which largely reflect the sensitivity threshold of current radial velocity programs, the
known extra-solar planets are a very diverse class. Some of
them have large eccentricities when others have nearly circular orbits, and their periods range between 3.3 days and
4.4 years. Giant planets can thus have very much shorter
periods than in our solar system, which clearly does not
represent the only possible outcome of planetary system
formation and evolution.
To date on the other hand, planets have mostly been
looked for around solar type stars, and, pulsar companions asides, they have only been found orbiting such stars.
Partly based on observations made at the Observatoire de
Haute Provence (CNRS)
⋆⋆
Partly based on observations obtained with the swiss 1.2m
telescope at the European Southern Observatory
Send offprint requests to: Xavier Delfosse, e-mail:
[email protected]
⋆
This reflects to some extent an understandable desire to
identify close analogs to our own solar system, which could
perhaps contain life sustaining planets. Also, the selection
function of the radial velocity planet searches has a relatively sharp optimum around spectral class G. Essentially
all stars hotter than approximately F5 have fast rotation
(Wolf et al., 1982), so that it is impossible to measure their
radial velocity to the ∼10m.s−1 accuracy needed to detect
planets. At the other end of the mass spectrum, most M
dwarfs have slow rotation (Delfosse et al., 1998a) and their
velocity can be measured accurately, as we discuss below.
Their luminosities however are much lower than those of
solar type stars. At a given distance a much longer integration time is thus needed to obtain a given radial velocity
precision on an M dwarf than on a G dwarf. All planet
search programs have thus understandably concentrated
on solar type stars.
G dwarfs however only represent a small fraction of the
disk stellar population, with the lower mass M dwarfs outnumbering them by about an order of magnitude (Gliese
& Jahreiss, 1991). It is thus likely that most planets in
our galaxy orbit stars whose mass and luminosity are significantly lower than the Sun’s (Boss, 1995), unless some
as yet unidentified physical process restricts planet formation to the environment of sufficiently massive stars. It is
clearly important to establish whether such a mechanism
exists.
For the last three years, we have been monitoring the
radial velocities of a nominally volume limited sample of
125 nearby M dwarfs. The two main goals of this large
observing program (∼30 nights/year) are to establish the
controversial (e.g. Kroupa, 1995, and Reid & Gizis, 1997,
for two contrasted views) multiplicity statistics of field M
dwarf systems, and to pin down the still uncertain massluminosity relation at the bottom of the main sequence.
Delfosse et al (1998b) present preliminary results for the
stellar companion search, with 12 new components found
in these nearby M dwarf systems, including the third detached M dwarf eclipsing binary (Delfosse et al., 1998c).
A byproduct of this program, related to the angular mo-
X. Delfosse et al.: The closest extrasolar planet
mentum dissipation of very low-mass stars, is described in
Delfosse et al (1998a).
Even though this was not the main focus of the program, we also realised from the start that for most of
these stars we obtain radial velocity precisions which are
sufficient to detect giant planets, if any exists around
them. We present in this letter the first such detection, around Gl 876 (BD−15◦ 6290, LHS 530, Ross 780,
HIP 113020), a V=10.2 M4 dwarf (Reid et al., 1995) at
d = 4.702±0.046 pc (ESA, 1997).
Delfosse et al. (1998a) present in detail the observed
sample, while Delfosse et al. (1998b) discuss the observing and analysis technique at length. We therefore only
briefly summarize this information in section 2. We then
proceed to discuss in section 3 the radial velocity detection of the planetary companion of GL 876. In section 4
we consider the implications of this detection and suggest
some possible follow-up observations.
2. Observing program
The sample contains the 127 M dwarfs listed in the third
edition of the nearby star catalog (CNS3 preliminary version, Gliese & Jahreiss, 1991) with a distance closer than
9 pc, a B1950.0 declination above -16 degrees, brighter
than V=15, and without a close much brighter primary.
Observations have been carried out since September 1995
with the ELODIE fiber-fed spectrograph (Baranne et al.,
1996) on the 1.93m telescope at Observatoire de Haute
Provence (OHP). The R=42000 spectra are wavelength
calibrated through simultaneous observations of a thorium lamp. Since June 1998 some southern stars have also
been observed with the nearly identical CORALIE spectrograph on the recently commissioned swiss 1.2m telescope at la Silla (Chile). CORALIE mostly differs from
the older ELODIE instrument by its spectral resolution
of R=50000, better sampling of the spectrograph PSF by
the CCD camera pixels, and significantly better temperature control. The first indications are that these modifications together result in a substantially improved intrinsic
stability (Queloz et al., in preparation).
The extracted M dwarf spectra are analysed through
cross-correlation with a binary (0/1) template constructed
from an observed ELODIE spectrum of Barnard’s star,
Gl699 (Delfosse et al., 1998b). For slowly rotating stars
the resulting velocities have internal standard errors (photon noise plus low level uncalibrated instrumental instabilities) which typically range from 10-15 m.s−1 for bright
−1
M dwarfs (V <
∼ 10) to ∼75 m.s at the magnitude limit
of the sample. For Gl 876 (V=10.2) typical standard errors are 10 to 20 m.s−1 , depending on airmass and seeing conditions. Magnetic activity is common in M type
dwarfs, and may further degrade the measurement accuracy (Saar et al., 1998). This potential error source is still
uncompletely characterised for M dwarfs, but for slowly
rotating stars (V sini < 3km.s−1 ) we can already bound
3
Table 1. Orbital elements of GL 876.
Element
Value
P
60.97
T
2450661.7
e
.336
ω
5.2
K1
248.0
V0
-1.902
a1 sin i
0.00131
f(m)
7.810−8
O-C(CORALIE)
23
O-C(ELODIE)
16
St. Err. Unit
.19
1.5
0.019
4.8
6.6
0.006
(rms)
(rms)
Days
Julian Days
deg
m.s−1
km.s−1
AU
M⊙
m.s−1
m.s−1
−1
it to σVr <
∼ 20m.s for our cross-correlation analysis with
the M4 binary template. Within the brighter two thirds of
the sample, a conservative assumption at the present time
is that we will detect all variables with semi-amplitudes
larger than 40m.s−1 . Assuming for illustration a 5 years
period, this corresponds to a 1 Jupiter mass (MJ ) planet
orbiting a 0.25 M ⊙ M4V primary (at 1.8 AU), or to a
2 MJ planet orbiting a 0.6 M ⊙ M0V primary (at 2.5 AU).
3. A planet around Gl 876
Since planet detection was not initially emphasized in the
observing program, its sampling strategy is not optimal for
detection of low amplitude variations on timescales shorter
than a few years. Gl 876 was observed once at each observing seasons in 1995 and 1996 and its velocity variations
became apparent from the three observations obtained in
late 1997. It was then marked in our program lists as a
variable. This low declination source however became unobservable from OHP before we could gather more data
and determine its orbit. The commissioning of the swiss
1.2m telescope at la Silla and its CORALIE spectrograph
in June 1998 provided the first opportunity to obtain 3
additional measurements of this southerly source, which
allowed to finally determine its orbit. These observations
were obtained within two weeks of the first light of this
telescope, providing an encouraging indication on its potential for planet discovery. An end of night measurement
from OHP at a large airmass provided a confirmation on
June 22, just in time to confidently announce the discovery
at the IAU “Precise Stellar Radial Velocities” conference
(Victoria, Canada, June 21st to 26th ). At this conference
we learned from G. Marcy that his group independently
identified the orbit of Gl 876, with orbital elements compatible with our own determination. Weather permitting,
we have since then attempted to observe Gl 876 at most
every three nights, and often every night. The 1998 data
therefore dominate the orbital solution.
The orbital solution is given in Table 1. Preliminary
solutions included a velocity zero point offset between the
northern (ELODIE) and southern (CORALIE) datasets
as a free parameter. The two systems were found to be
4
X. Delfosse et al.: The closest extrasolar planet
Fig. 1. Combined ELODIE and CORALIE radial velocities for Gl 876. The solid line is the radial velocity curve
for the orbital solution.
entirely consistent, and this parameter was thus held fixed
to zero for the final solution. Figure 1 shows the individual radial velocity measurements as a function of orbital
phase (the 16 orbital periods elapsed since the first measurement make unpractical a display as a function of time;
we however have essentially continuous coverage of one period in June and July 1998, excluding any possible spectral
alias). The orbital period is two months and the velocity
semi-amplitude is ∼250 m.s−1 , over 10 times the standard error of one radial velocity measurement. The radial
velocity curve implies a moderate but highly significant
eccentricity of e=0.34.
The large amplitude and moderate period of the radial velocity variation argue strongly for orbital motion as
its cause. An integration of the radial velocity curve implies a minimum physical motion of ∼0.5 R⊙ . This variation is about twice the radius of an M4 dwarf (Baraffe &
Chabrier, 1996; Chabrier & Baraffe, 1997), excluding pulsation as a possible explanation. Gl 876 in addition only
has low level photometric variability, and indeed happens
to be one of the standard stars of the original UBV system (Johnson & Harris, 1954). It is actually variable, but
with low rms amplitudes of 13 mmag at V, 9 mmag at
R and 6 mmag at I (Weis, 1994). The photometric variations don’t phase well at 61 days and appear consistent
with a BY Dra type variability (Sasselov and Cody, private communication). Since Gl 876 is a very slow rotator
(Vsin i <2km.s−1 , Delfosse et al., 1998a), rotational modulation of such low level surface inhomogeneities cannot
explain its large radial velocity variations. Densely sampled photometry would on the other hand be of considerable interest to establish the stellar rotation period.
The mass of Gl 876 unfortunately contributes some
uncertainty to the minimum mass of its companion. As a
consequence of H2 recombination in the photosphere and
the deepening convection for lower mass stars (Kroupa et
al. 1990), the luminosity does not drop nearly as quickly
per unit mass for mid-M dwarfs as it does for both higher
and lower mass stars (Henry & Mc Carthy, 1993), and it
has a stronger metallicity dependence. Between ∼0.50 M ⊙
and ∼0.18 M ⊙, Mass-Luminosity relations therefore have
both shallow slopes and large intrinsic dispersions (Henry
& Mc Carthy, 1993). Gl 876 thus belongs to a spectral
type range where the mass of a single star is poorly
constrained by its observable characteristics. Taking at
face value either the solar neighborhood observational
mass-luminosity relation of Henry & Mc Carthy (1993)
or the solar metallicity models of Baraffe et al. (1998),
the absolute magnitudes of Gl 876 (MV =11.81, MJ =7.56,
MH =6.96, MK =6.70, Leggett (1992) and ESA (1997)) imply a mass of 0.30±0.05 M ⊙ for Gl 876. As an illustration
of possible uncertainties however, Delfosse et al (1998c)
measure M = 0.432 ±0.001 M ⊙ and MV = 11.7±0.2
(MV =11.81 for Gl 876) for the brighter star in GJ 2069A,
an M3.5V eclipsing binary which is probably super-metalrich ([M/H]∼+0.5). From its position in colour-colour diagrams (Leggett, 1992), and from the relative depth of
its cross-correlation dips with several binary templates
(Delfosse et al., in preparation), Gl 876 is probably more
metallic than the sun, though not as much as GJ 2069A.
We adopt a mass of 0.3 M ⊙ for the rest of the discussion
but warn that it is uncertain by perhaps 30%. The minimum semi-major axis and planetary mass which result
from the orbital solution are then asin i = 0.20 AU and
2
1
M2 sin i = 2.0MJ . They respectively scale as M 3 and M 3 .
4. Discussion
The orbital elements of Gliese 876b are worth noting. In
spite of the low mass of Gl 876, the ice-condensation radius
at the time of planet formation around this star is ∼4 AU,
only 20% lower than around a solar type star (Boss 1995).
Once again the orbital semi-major axis (a = 0.20 AU) is
thus much smaller than the expected minimum radius for
giant planet formation, and some orbital migration must
have occcured. However the observed orbital separation is
also 4 times larger than the measured semi-major axes of
51 Peg, τ Boo and υ And (0.04-0.05 AU). The excess of
planets with such small semi-major axes is believed to result from outward torques which counteract at short distances the inward torque induced by the interaction of
the planet and the protoplanetary disk (Lin et al. 1996,
Trilling et al. 1998). These torques only become effective
at separations significantly smaller than 0.20 AU, and cannot have played a signficant role for the Gl 876 system.
It is also interesting to note that the orbit of GL 876b
is eccentric (e=0.35), while interaction with an accretion
disk is expected to damp any significant orbital eccentricity of a planet (Goldreich and Tremaine, 1980). Several mechanisms may be called in to explain the large orbital eccentricities of giant extrasolar planets, but in the
present case the most interesting possibility is probably
the chaotic interaction of several giant planets (Weidenschilling & Marzari, 1996; Rasio & Ford, 1996; Lin & Ida,
1997). A frequent final result of such a strong gravitational
X. Delfosse et al.: The closest extrasolar planet
interaction is a planetary system with a single giant planet
at a moderate semi-major axis, in an eccentric orbit. This
could thus simultaneously explain the semi-major axis and
the eccentricity.
Contrary to all previously confirmed planets around
main sequence stars, Gl 876b orbits a star which is very
different from our Sun, showing that planetary systems
form around stars of widely different types. Gl 876 is
much less massive than the Sun, ∼0.3 M ⊙, and at most
only ∼150 times more massive than its planet. Its radius is only three times as large: the radius of Gl 876 is
∼0.3 R⊙(Chabrier & Baraffe, 1997), while that of Jupiter
is 0.10 solar radii. Gl 876 is also much cooler than the Sun,
and much less luminous. From the observed I-K and V-I
colours (Leggett, 1992) its effective temperature is 3100
to 3250 K (Leggett et al., 1996), compared with 6000 K
for the solar surface. From its absolute V magnitude and
the bolometric correction of Delfosse et al. (1998a) it bolometric magnitude is 9.42, corresponding to 1.35 10−2 L⊙ .
Even though the planet of Gl 876 is twice closer to its star
than Mercury is to the Sun, the stellar flux at Gl 876b is
only ten times the solar flux at Jupiter, and lower than
the solar flux at Mars. The apropriate albedo for Gl 876b
is unclear, and, by analogy with Jupiter (e.g. Podolak et
al., 1993), its thermal balance of Gl 876b may also include
a substantial contribution from an internal heat source. A
detailed evaluation of its effective temperature is thus beyond the scope of the present letter, but Gl 876b is clearly
much too cold to possibly sustain liquid water above the
1 bar level.
Gl 876 is closer to us than all other stars orbited by
known extra-solar planets, by at least a factor of 3. At
d=4.7 pc, Gl 876 is the 40th closest stellar system to our
Sun, and the 53rd closest star. Since M dwarfs make up
∼80% of the solar neighbourhood population (Gliese &
Jahreiss, 1991), it is only natural that the first member of
this numerous class found orbited by a planet is a very
nearby one, unless planetary formation would have selected against low mass stars. This discovery weakens such
an hypothesis but improved statistics would obviously be
needed for a reliable conclusion.
This detection represents an opportunity to confirm
a radial velocity detected planet through astrometry and
determine its actual mass, or at the very least to set a
lower limit which is firmly planetary. Gl 876 is both at
least 3 times closer to us than any other star with a detected planetary companion, and about 4 times less massive (only 0.3 M ⊙ instead of about ∼ 1 M ⊙ for all previous detections). Despite its relatively short orbital period
of 61 days, the astrometric reflex motion induced by its
∼2MJ companion is therefore unusually large by extrasolar planet standards, with a minimum semi-major axis
of 0.27 milliarcsecond for an edge-on orbit, and larger
by 1/sin(i) for more face-on geometries. The best single
measurement precision of an astrometric observations is
at present 1 milliarcsecond, with the FGS instrument on
5
HST. A detection is clearly an ambitious measurement at
this time if the orbit is seen edge-on, and it would need
a very determined effort. A lower limit on the inclination
of | sin i| > 0.25 on the other hand will be easily obtained,
and would already imply M2 <8 MJ . In addition, these
observations can be accomplished over the short timescale
of one orbital period, only 2 months.
Finally, it is interesting to note that the measurements
of Gl 876 obtained with the new CORALIE spectrometer on the 1.2 m telescope have residual O-Cs as small
as 22 m/s, for a V magnitude of 10.2. This discovery
of a giant planet around a rather faint M4 dwarf illustrates that this small telescope will powerfully contribute
to the search for extrasolar planets. The application of the
cross-correlation technique to the full wavelength domain
(300 nm) of CORALIE compensates to some extent the
disadvantage of the small telescope aperture (Baranne et
al. 1996), and the nearly full-time availability of the telescope for planet searches will make future period identifications much easier.
Acknowledgements. We thank the technical staff and telescope
operators of OHP for their support during these long-term observations. We thank Claudio Melo for obtaining additional
measurements for us with CORALIE, and the technical teams
of the Geneva and Haute Provence Observatories for building
this excellent instrument, in particular Luc Weber for his hard
and efficient work on the CORALIE software. We are grateful to the referee, Gilles Chabrier, for his constructive comments, and to Dimitar Sasselov for pointing out to us the low
level photometric variability of Gl 876. X.D. acknowledges support by the french Ministère des Affaires Etrangères through
a “Lavoisier” grant.
References
Baraffe I., Chabrier G., ApJ 461, L51.
Baraffe I., Chabrier G., Allard F., Hauschild P., 1998, A&A,
in press.
Chabrier G., Baraffe I., 1997, A&A 327, 1039.
Boss A.P., 1995, Science 267, 360.
Delfosse X., Forveille T., Perrier C., Mayor M., 1998a, A&A
331, 581.
Delfosse X., Forveille T., Beuzit J.L., Udry S., Mayor M., Perrier C., 1998b, submitted to A&A.
Delfosse X., Forveille T., Mayor M., Burnet M., Perrier C.,
1998c, submitted to A&A.
ESA, 1997, The Hipparcos Catalog, ESA SP-1200.
Goldreich P., Tremaine S., 1980, ApJ 241,425.
Henry T.J., Mc Carthy D.W., 1993, AJ 106, 773.
Johnson H.L., Harris D.L., 1954, ApJ 120, 193.
Kroupa P., Tout C.A., Gilmore G., 1990, MNRAS 244, 76.
Kroupa P., 1995, ApJ 453, 358.
Gliese W., Jahreiss H., 1991, Preliminary Version of the Third
Catalog of Nearby Stars, as available at CDS Strasbourg.
Leggett S.K., 1992, ApJSupp. 82, 351.
Leggett S.K., Allard F., Berriman G., Dahn C.C., Hauschild
P., 1992, ApJSupp. 82, 351.
Lin D.N.C., Bodenheimer P., Richardson D.C., 1996, Nature
360, 606.
6
X. Delfosse et al.: The closest extrasolar planet
Lin D.N.C., Ida S., 1997, AJ 477, 781.
Mayor M., Queloz D., 1995, Nature 378, 355.
Marcy G.W., Butler R.P., ARAA 36, 57.
Podolak M., Hubbard W.B., Pollack J.B. 1993, in Protostars
and Planets III, ed. E. H. Levy, & J. I. Lunine (Tucson:
Univ. Arizona Press), 1109.
Queloz D., 1999, ”The new planetary systems”, in “Planets
outside the solar system”, NATO Advanced Study Institute
school held at Cargèse, France, J. M. Mariotti (ed.), in
press
Rasio F.A., Ford E.B., 1996, Science 274, 954.
Reid I. N., Hawley S. L., Gizis J. E., 1995, AJ 110, 1838.
Reid I. N., Gizis J. E., 1997, AJ 113, 2246.
Saar S.H., Butler R.P., Marcy G.W., 1998, ApJ 498, L153.
Trilling D.E., Benz W., Guillot T., Lunine J.I., Hubbard W.B.,
Burrows A., 1998 ApJ 500, 428.
Weidenschilling S.J., Marzari F., 1996, Nature 384, 619.
Wolf S.C., Edwards S., Preston G.W, 1982, ApJ 252, 322.